Usage Dashboard
The Usage Dashboard gives you a detailed view of how Ileen’s AI agents are being used on your account. Open it any time from the sidebar footer to understand costs, spot heavy workloads, and plan your credit consumption.
Opening the dashboard
Click your avatar or the Usage link in the sidebar footer. The dashboard opens as a modal overlay.
Selecting a time period
Use the period selector in the top-right corner:
| Option | What it shows |
|---|---|
| Last 7 days | Rolling 7-day window |
| Last 14 days | Rolling 14-day window |
| Last 30 days | Rolling 30-day window (default) |
| Last 90 days | Rolling 90-day window |
| All time | Every request on your account |
| Custom range… | Opens a date-picker pair; choose any start and end date |
The dashboard reloads automatically when the period changes.
KPI cards
The top section shows at-a-glance metrics for the selected period:
| Card | What it measures |
|---|---|
| Total Requests | Number of AI agent calls made |
| Total Cost | Sum of all LLM costs in euros (€) |
| Total Tokens | Combined input + output tokens; breakdown shown below the number |
| Cache Tokens | Tokens served from the prompt cache (read + write); only shown when non-zero |
| Avg Duration | Mean time per agent call (shown as ms / s / min) |
| Success Rate | Percentage of requests that completed without error |
| Projects | Number of distinct projects that triggered agent calls |
Daily Cost & Requests chart
The area chart plots two series over time:
- Cost (€) — left Y-axis, purple area
- Requests — right Y-axis, cyan area
Hover over any date to see exact values in the tooltip. The chart is hidden when there is only one data point.
Breakdown tables
Below the chart, two side-by-side tables break down usage further:
By Agent
Shows which agent type consumed the most resources. Each row contains:
- Agent — the agent name (e.g.,
Architect,TechLeader,TeamLeader,CodeAnalyst) - Requests — number of calls
- Tokens — total tokens consumed
- Cost — cost in euros
By Operation
Shows the same metrics grouped by operation type (e.g., generate_plan, chat, code_query, deploy). This helps identify which specific tasks are most expensive.
Practical use cases
Budget monitoring — Set the period to “Last 30 days” and compare Total Cost to your credit balance to estimate when you’ll need to top up.
Finding expensive agents — Sort the By Agent table mentally by cost; if TeamLeader is dominant, you may be triggering too many task regenerations.
Project profiling — Note the Projects KPI. If one project accounts for most of the cost, consider whether its briefing or task descriptions can be made more efficient.
Troubleshooting low success rate — A success rate below 95% suggests recurring agent failures. Cross-reference with the Monitoring Active Jobs guide to identify stuck or erroring jobs.